Little figures behind the big figures
June 2, 2009
When you look at most polls, it is the figures behind the headlines which are the most important – chief among these is sample size.
If you’re carrying out a poll which aspires to anything beyond than the status of the anecdotal, the reader needs an idea of its scale (clearly, other measures such as margin of error are helpful, too). This is really the minimum requirement of credibility.
The Liverpool Daily Post’s local government blog makes interesting points about Merseyside reaction to the MPs’ expenses scandal, but the survey it draws on is deeply unsatisfactory.
“Poll reveals trust in MPs has fallen by half” says the headline, backed up by detail that prior to the scandal, 50% of respondents trusted MPs, but now the figure is only 28%. We also learn that, understandably, two in three of the survey participants have been angered by the affair.
However, without knowing how many took part in the survey, the results are pretty meaningless. Presumably, the figure wasn’t quoted in the article because it was on the low side, but knowing the size of the sample is essential if readers are to draw any conclusions.
Even if it turns out only a couple of dozen Merseysiders responded, the story is still of interest – but at least armed with this fact, its significance can be evaluated. It is a story on an issue of great significance to the region, so it’s a pity the lack of this little figure detracts from the bigger picture.
One touch I did like was the wordcloud which highlighted words left in user comments, with the size of the word reflecting its popularity (similar to a tag cloud) – it’s a nice feature. Word clouds like this can be generated by sites such as Wordle.